Shift coke mesh center to gravity center

[JimmyDaSilva]

According to @nlyubova, the mesh center should be it's gravity center.
Making the shift solve my orientation problem

[vrabaud]

Well, that would now have the tabletop tutorial fail right ?
@nlyubova, this behavior should be documented: https://github.com/wg-perception/linemod/blob/master/doc/source/index.rst
The center of gravity is not an easy thing to compute are you sure this is what it's supposed to be ? Shouldn't it be the center of the bounding box ? Please show me in the code where that matters.

[JimmyDaSilva]

@vrabaud yes that will fail the tabletop. It make sense to me now why you had the mesh shifted.
Though the initial center was not on the axe of the cylinder, but on one corner of the bounding box.
I'm going to check quickly the results with a mesh center on the cylinder's base with linemod.
I have no clue how the computation is linemod works. I was following @nlyubova 's advices.

[JimmyDaSilva]

Seems fine. And it will work with both linemod and tabletop. I will a new PR
The green real_icp_model is still shifted though

[JimmyDaSilva]

@vrabaud It should be fine now

[vrabaud]

ok, just before merging, @nlyubova, how come that was not done before ? Has code changed ?

[nlyubova]

But for all our meshes, the center is already the center ! The cola mesh from the tutorial is the only object with a center shifted to the bottom. So if we change the code, all our meshes won't work

[nlyubova]

@vrabaud as you asked, I've added details about the mesh center to https://github.com/wg-perception/linemod/blob/master/doc/source/index.rst